Every year someone declares affiliate marketing dead, and every year the spend goes up. In 2026 both halves of that sentence are true at once, which is the thing worth understanding. US advertisers will spend an estimated 13.81 billion dollars on affiliate marketing this year, up 11.3 percent on 2025 and roughly double the 6.7 percent growth rate of US retail ecommerce overall, according to eMarketer. A channel growing at twice the rate of the market it serves is not dying.
And yet the people who built businesses on affiliate marketing are not imagining their pain. Content review sites are losing traffic by the half. Amazon quietly cut commission rates by as much as 50 percent. Coupon extensions got caught skimming credit from the creators who earned it. All of that is real.
The mistake is treating these as one story. They are not. Affiliate marketing is unbundling: the original product, a website that earns a cut for sending a buyer to a merchant, is splitting into a part that is contracting hard and a part that is growing fast. The word "affiliate" used to name one thing. It now names at least four, and the gap between them is the whole point.
Where the word came from
Affiliate marketing started as a fix for a narrow problem: how do you pay a website for a sale you cannot see it produce. The answer was a tracked link and a cookie. William J. Tobin ran an early revenue-share program for PC Flowers and Gifts and filed a US patent on affiliate tracking in January 1996. CDNow let music sites link to album purchases for a cut. Then Amazon opened its Associates program in 1996, and because Amazon was widely copied, the open affiliate program became the default. Anyone with a website could join, drop a link, and get paid per sale.
For two decades that model worked because of one assumption: people found products by searching, reading, and clicking. A buyer would search "best noise-cancelling headphones," land on a review site, read a comparison, click a link, and buy. The review site earned its commission for doing the genuine work of helping someone decide. Networks like Commission Junction, LinkShare and ShareASale built the tracking and payment plumbing so merchants did not have to.
That assumption is what is now breaking. Not the idea of paying partners for performance. The specific assumption that a search-and-click journey is how discovery happens.
What is shrinking, and why
Three pressures are squeezing the classic content-and-coupon affiliate at the same time, and they compound.
The first is AI search. When Google answers a query inside an AI Overview, the click that affiliate tracking depends on often never happens. For news queries, zero-click searches rose from 56 percent to 69 percent between May 2024 and May 2025, according to Similarweb. Google search traffic to publishers fell by roughly a third globally in the year to November 2025, on Chartbeat data AdExchanger reported. Review sites took it hardest. The New York Times' Wirecutter lost somewhere between 65 and 72 percent of its Google search visibility depending on the dataset, with sharp drops in mid-2025, according to an analysis by Glenn Gabe. When the traffic halves, the commission halves with it.
The second is commoditized content. AI made it nearly free to produce a passable product roundup, so the supply of affiliate review content exploded while its average quality fell. The market was already concentrated before that. When the SEO analyst Glen Allsopp studied 10,000 product-review search results, only 4 of the top 100 ranking domains were independent sites; the rest belonged to news organizations, public companies, or large networks. An independent reviewer is now squeezed from above by media conglomerates and from below by infinite cheap content. Ahrefs published a much-discussed piece arguing it would not start a pure affiliate site today, and the argument is hard to dismiss.
The third is margin compression at the merchant. Amazon overhauled its Associates program with no public announcement, cutting commission rates by as much as 50 percent, removing milestone bonuses, and degrading reporting down to less useful tracking-ID-level data. The changes reached the US around early March 2026, Adweek reported, and multiple publishers described it as part of a directive to cut program costs by 20 percent. One deal-site publisher told Hello Partner it expects its 2026 Amazon revenue to come in half of what it had projected. When the largest program in the channel cuts payouts and stops explaining itself, every smaller program feels licensed to do the same.
The coupon and cashback corner has its own version of the squeeze. Discount and promotion publishers took 42.4 percent of US affiliate revenue in the first half of 2025, up from 39.7 percent a year earlier, per Awin data reported by eMarketer. That sounds like growth, but it has reopened an old argument about whether those publishers create demand or just intercept it at checkout. The Honey scandal made the argument unavoidable. A December 2024 investigation by the YouTuber MegaLag alleged the PayPal-owned coupon extension overwrote affiliate cookies at the final click, so that under last-click attribution Honey collected commissions on sales other partners had originated. Class-action complaints followed within days, and PayPal has contested them. Whatever the courts decide, the episode forced brands to ask a question they had avoided: which of our affiliates actually caused a sale, and which just stood near the cash register.
So the shrinking part is specific. It is the affiliate whose business is ranking a review in Google, or whose business is capturing credit at checkout. Both depended on a funnel that AI is compressing and an attribution rule that is losing legitimacy.
What is growing, and why it counts as the same channel
Here is the part the "affiliate is dead" headlines miss. While the content-and-coupon model contracts, a wider category is expanding under the same budget line, and the industry has a name for it: performance partnerships.
Start with creators. On Awin's network, content creators' share of affiliate revenue rose from 15.9 percent to 19.5 percent year over year, making them the fastest-growing publisher type, per eMarketer. A creator partnership is structurally an affiliate deal, a tracked link and a performance payout, but it does not depend on Google rankings. It runs on an audience the creator owns, and it pays out the same way whether the buyer found the product in a video, a newsletter, or a story. The same eMarketer reporting notes that repeat creator integrations compound: by the eighth integration with the same YouTube creator, affiliate link clickthrough reached 1.8 times the rate of the first.
Then there are the partner types that were never "affiliates" in the old sense at all. B2B referral partners, paid per qualified subscription rather than per cart. Brand-to-brand tie-ups, where two non-competing companies with shared audiences cross-promote and split the upside, which Awin lists among its fastest-growing models. Card-linked offers, where cashback attaches to a registered payment card instead of a click or cookie. None of these uses the search-and-click journey that AI is eroding, and all of them sit on the same partnership platforms and inside the same budget.
This is why the consolidation news matters. On 28 April 2026, Rakuten and impact.com announced an alliance: Rakuten migrates the merchants and publishers on its partnership-marketing tech onto impact.com's platform, and refocuses on managed services and its Rakuten Rewards cashback business. AdExchanger reported it as two of the longest-standing rivals in the industry choosing to combine rather than compete. Read it through the unbundling lens and it is straightforward. The technology layer, where the action is, consolidates around platforms that manage every partner type at once. The managed-service layer separates out as its own business. The thing being unbundled is the old all-in-one network itself.
The label "performance partnerships" is not just rebranding. It is an admission that "affiliate" stopped describing one job. A coupon extension, a B2B SaaS referrer, a YouTuber with a long-running brand deal, and a thin review site all got paid through the same pipes, so they all got called the same name. They were never the same business. Unbundling is the market finally pricing them differently.
Where this goes next
Two forces will decide how the split settles.
The first is measurement. Last-click attribution, the rule that lets the partner nearest checkout claim the sale, cannot survive a world where the click happens late or not at all. The Honey case made the unfairness visible; AI search makes the rule simply inaccurate, because the influence that drove a purchase increasingly happens inside a chat answer with no click to record. Expect brands to move toward multi-touch views and incrementality testing, which asks the harder question: which partners bring customers who would not have converted anyway. That question is bad news for the intercept-at-checkout model and good news for partners who genuinely create demand. Tracking is also moving server-side and first-party, because third-party cookies are no longer dependable enough to run a payout on.
The second is agentic commerce, and it is genuinely unsettled. As buyers ask AI assistants to compare options and find deals, affiliate content turns out to be what those assistants read. eMarketer cites a striking figure: when ChatGPT discussed the eyewear brand Zenni, close to 70 percent of the sources it drew on were affiliate content, while brand-owned sites make up only a small fraction of what AI surfaces. By that measure affiliate content is the substrate of AI product discovery, which would make affiliates more important, not less.
But importance is not the same as getting paid. The open question is whether a tracked, commissionable link survives an agent-mediated purchase, or whether platforms route around partners and monetize through their own fees and ads. The early signals are mixed. OpenAI launched Instant Checkout inside ChatGPT in September 2025, then in March 2026 pulled back to a discovery-first model, letting merchants keep their own checkout while it focuses on product discovery. Only about 30 Shopify merchants had gone live, and analysts told CNBC that OpenAI underestimated the work of onboarding merchants, showing accurate product data, and supporting multi-item carts and loyalty memberships. Discovery-first is, for now, the affiliate-friendly outcome: the assistant researches, the merchant keeps checkout, and the path a partner influenced is still traceable. Whether it stays that way is the single biggest variable hanging over the channel.
For brands building agentic commerce experiences, this is a live design decision, not a far-off one. If an AI assistant or a brand's own agent guides a buyer through research and toward a purchase, the system has to record which partner shaped that journey, or the performance-partnership model quietly stops working. Designing agent flows so that partner attribution survives the handoff to checkout is the kind of plumbing problem that decides whether the growing half of the channel keeps growing. It is the sort of work Perform Digital does when it helps enterprises put agents into a commerce stack: making sure the measurement holds up when the funnel no longer looks like a funnel.
The practical takeaway is to stop asking whether affiliate marketing is dead and start asking which affiliate marketing. A brand whose program is mostly thin content sites and checkout-intercept coupons owns a shrinking asset and should expect it to keep shrinking. A brand building always-on creator partnerships, B2B referral relationships, brand-to-brand tie-ups, and card-linked offers, measured by incremental contribution rather than last click, owns a part of the channel that is growing faster than ecommerce itself. Both are called affiliate marketing. Only one of them describes what the next few years will reward.
Council summary
This post argues that affiliate marketing is not dying but splitting in two: a content-and-coupon model squeezed by AI search, commoditized review content, and merchant commission cuts, and a wider performance-partnership model built on creators, B2B referrers, brand tie-ups, and card-linked offers that is growing faster than ecommerce. The council verified every figure against primary reporting, including the eMarketer 13.81 billion dollar 2026 forecast, the Amazon Associates cuts, the Awin creator and coupon share data, the Wirecutter visibility drop, and the OpenAI Instant Checkout pullback. Four claims were corrected: the 56 to 69 percent zero-click figure was qualified as news queries only, the OpenAI merchant count was fixed from "around a dozen" to roughly 30 with the reasons OpenAI actually gave, an unverified "majority of brands" budget-shift claim was cut, and the Rakuten framing was corrected since the cited article does not call it the first network-to-network deal. The takeaway for a brand is to stop asking whether affiliate marketing is dead and audit which half of it the program actually owns.
Comments